DAILY NEWS AND INFORMATION FOR THE GLOBAL GRID COMMUNITY / SEPTEMBER 8, 2003: VOL. 2 NO. 36    ( Table of Contents )

Special Features:

GRID COMPUTING -- THE TRUTH BEHIND THE HYPE
By Alan Beck, Editor-in-Chief

SAS has become the newest member of the Global Grid Forum. GRIDtoday talks with Tho Nguyen, SAS manager of data integration strategy, about Grid computing-- what it means to the industry, to SAS and to its customers.

GRIDtoday: Why is Grid computing such a hot topic? What benefits does it offer?

NGUYEN: The dilemma is that IT shops are faced with declining budgets and therefore need to more fully use the servers and workstations they already have available within the enterprise. Organizations are inundated with data, but they need to process it to obtain useful intelligence for business decisions. The workload to process the data is overwhelming and continues to increase at an alarming rate. Grid computing addresses this dilemma.

Grid computing lowers costs and boosts efficiencies of computing networks by optimizing resources, distributing workloads and decreasing data processing time. Computationally intensive projects that would typically take weeks or months to process can now be completed in days. Grid computing makes it possible to get quicker results while reducing the total cost of ownership.

GRIDtoday: What role does SAS play in Grid computing?

NGUYEN: SAS has been involved with Grid computing for quite a while through our advanced parallel processing software. We have formalized our involvement by becoming a sponsor in the Global Grid Forum. We have customers from a variety of industries that are using SAS technology in a Grid environment to reduce total elapsed execution time on a project by 92 percent to 99 percent. An example is the National Institute of Environmental Health Sciences (NIEHS), which uses SAS software to help its scientists achieve a 95 percent reduction in total elapsed execution time on key research projects. Additional information on SAS support for Grid computing is available at support.sas.com/rnd/scalability/Grid/index.html.

Our involvement with Grid computing will continue to grow. We'll offer more updates in the near future.

GRIDtoday: Why is this good news for SAS customers?

NGUYEN: SAS enables customers to do more with less, ultimately lowering the cost of IT ownership. Customers can leverage Grid computing to optimize unused capacity within a network. Grid computing allows our customers to handle their exploding workload requirements to get the information they need in a reasonable timeframe while also staying within IT budgets.

Our participation in the Global Grid Forum keeps SAS at the forefront of this new movement, helping our customers get the computing power and network bandwidth they need to perform critical tasks that improve their businesses. Our involvement also enables us to exchange knowledge and information with industry leaders in order to advance our current Grid capabilities, as well as to help shape the Grid-computing industry in general. This is all good news for our customers.

GRIDtoday: How does Grid computing work?

NGUYEN: Grid computing is a method of harnessing the power of many computers in a network to solve problems requiring a large number of processing cycles and involving huge volumes of data. Grid computing taps into the unused processor capacity of hundreds, sometimes even thousands, of computers to help users achieve much faster results on large projects at a lower cost.

Computing Grids are basically networks that pool resources -- CPU cycles, storage or data -- from many different machines and apply them to one complex application. The networks are referred to as computing Grids because they mirror the behavior of electrical Grids. When you turn on your light, you're not drawing electricity from a single power plant, but rather you are accessing a network of power plants spread throughout the region. Grid computing operates in the same way -- allocating processing power wherever it is needed.

GRIDtoday: How far along are most of your customers in terms of implementing Grid?

NGUYEN: While the NIEHS and others are already reporting significant gains, Grid computing is still in its infancy. Many of our customers are just starting to understand the true value of Grid computing. As standards and enabling technologies continue to evolve, more of our customers will be able to create a sophisticated Grid-computing environment.

GRIDtoday: What are some of the issues users face when implementing Grid computing into an enterprise?

NGUYEN: Grid computing is an emerging technology and a standard approach has not been established. Without a standard approach, the process is complex and customization is required. Customers often seek assistance from the vendor's professional services to configure and integrate the architecture. In the future, turnkey solutions will be available in the marketplace and customers who plan to implement Grid computing will benefit from off-the-shelf applications.

GRIDtoday: What does the infrastructure of Grid computing look like?

NGUYEN: There are multiple layers in Grid computing, with the higher layers being more software-centric and the lower layers more hardware-centric. Starting from the top, you have the application and services layers, which consist of portals and development toolkits. Next is the middleware layer, which provides the protocols that enable multiple elements (i.e., servers, storage, networks) to participate in a unified Grid environment. The resource layer is next and is made up of the actual resources that are part of the Grid, primarily servers and storage devices. The last layer is the network layer, which is the underlying connectivity for the resources in the Grid and includes switches, routers, et cetera.

GRIDtoday: What industries are most interested in Grid computing?

NGUYEN: Academic research and health/life sciences are traditionally the two areas where Grid computing has been focused because of the volumes of data being processed. Grid computing is being considered in the other vertical industries that require enormous amounts of compute power, complex computer processing cycles or access to large data sources to effectively manage the workload. These industries include manufacturing (computer and industrial), financial services and government.

According to Insight Research, Grid computing is one of several fundamental technologies that many predict will transform the IT industry and the enterprise IT infrastructure over the next decade. In addition, Grid Technology Partners -- an analyst firm specializing in Grid computing -- has said that the overall Grid-computing market could grow to $4.1 billion by 2005.

GRIDtoday: What are the components of Grid computing?

NGUYEN: At SAS, we believe that a complete Grid solution will incorporate features such as load balancing, job scheduling and restart capabilities, and security, as well as an intuitive user interface for load balancing, job scheduling and resource allocation.

GRIDtoday: How will Grid computing become more widely accepted by businesses since it is such a technical concept?

NGUYEN: New standards, enabling technologies, toolkits and products are becoming available that allow businesses to use and appreciate the advantages of Grid computing. In general, Grid computing implies support for multiple platforms and platform transparency, which are critical to a business being able to adopt new technologies.

GRIDtoday: What other vendors are involved in Grid computing, and how does SAS differentiate itself from these other companies?

NGUYEN: Other key Grid-computing players include IBM, Oracle, Sun Microsystems, Hewlett-Packard, Globus, Avaki, DataSynapse, Entropia, Platform Computing and United Devices. Note that IBM, Platform Computing, HP and Sun are partners of SAS. There are a number of other companies that have recently begun their own Grid-computing initiatives and have joined the Global Grid Forum. Check out www.ggf.or g/L_Involved_Sponsors/2003_spons.htm for a comprehensive list of Global Grid Forum sponsors.

In some ways, Grid computing is like a puzzle that requires pieces from different vendors in order to work. Each piece is critical and affects how well a Grid will perform. SAS' contribution lies in our parallel processing capabilities. What we are announcing is just the tip of the iceberg in terms of what we plan to offer our customers who need a Grid-computing environment. And we are unique because we are the only vendor to deliver a truly complete portfolio of business intelligence technologies that our customers can use as needed -- allowing them to extend the benefits of a Grid-computing environment.

GRIDtoday: Can you give some "real world" examples of how Grid computing works?

NGUYEN: Here are a few examples:

1. The NIEHS has set up a Grid environment to help with its research in finding a cure for cancer. Scientists there are using their computing resources to more quickly handle statistical analysis computations. In turn, this helps the scientists examine their data much faster than they could before. Prior to Grid, it would take them months or years to process and access all the information available to them, but now they are able to achieve a 95 percent reduction in total elapsed execution time on key research projects.

2. A large pharmaceutical company has implemented a Grid environment to process sales data in 600 territories. The IT department was processing large volumes of data that took more than 26 hours and struggled to deliver information quickly. Since they have implemented a Grid, processing 600 sales territories took less than two hours -- a 93 percent reduction in time.

3. Farther down the line, once Grid computing is more established and has extended beyond businesses to an individual user, Grid computing will deliver an Internet that performs much faster with much higher availability and better security. For example, how often have you tried to visit a Web site and it was down or too busy? With Grid computing, redundancy will be built in with multiple copies of Web sites available to ensure that you get the information you are looking for on the first visit. The system will make decisions to route you to the site that can most quickly handle your request -- similar to what happens when you turn on your light switch. You don't care where the power is coming from, but just that you have electricity.

GRIDtoday: What is the OGSA?

NGUYEN: OGSA stands for Open Grid Services Architecture -- a standard on which next-generation Grid applications will be based. The Global Grid Forum has a working group, the Open Grid Service Infrastructure, which reviews and refines the Grid Service Specification and other documents that derive from this specification, including OGSA infrastructure-related technical specifications and supporting informational documents.

GRIDtoday: What is the difference between Grid computing and utility computing?

NGUYEN: Those terms are commonly confused -- as is "on-demand" computing, which is primarily an IBM term for Grid computing. Grid computing is a collection of resources used together to solve a common problem. Utility computing primarily involves a much wider public domain than Grid computing (although it can also be deployed internally through a company's IS department). Grids are not necessary for utility computing, but they can be part of it. In general, utility computing requires a service provider (for example, an ASP or hosting service) that offers on-demand services on a pay-as-you-go basis. While Grid computing involves only the hardware and software necessary to deploy a Grid environment, utility computing is more focused on ensuring that the user sees the desired results. As the technology involved in Grid computing and utility computing continues to mature, we'll hear more about how the two initiatives interrelate.

GRIDtoday: What is the Globus Project? Is SAS involved?

NGUYEN: It is an R&D project focused on enabling Grid applications by providing tools that make it easier to develop computational Grids. The Globus Project facilitates the development of the Globus Toolkit, which is central to hundreds of science and engineering Grid deployments worldwide, and it is increasingly important to corporate strategies for overcoming obstacles to collaboration. The Globus Project is leading the effort to define the OGSA. Stay tuned for more details about SAS' involvement with the Globus Project.

GRIDtoday: How do Grid computing and Web services relate?

NGUYEN: The two concepts are definitely coming together. The purpose of the OGSA is to help merge Grid-computing protocols with Web-services protocols and use Web-services standards like SOAP (Simple Object Access Protocol); WSDL, which is the XML-based standard for exchanging content; and other standards for services.

GRIDtoday: What makes Grid computing cost-effective?

NGUYEN: Grid computing can be a cost-effective way to resolve IT issues in the areas of data, computing and collaboration, especially if they require enormous amounts of compute power, complex computer processing cycles or access to large data sources. We also believe that Grid computing needs to be a secure, coordinated sharing of heterogeneous computing resources across a networked environment that allows users to analyze their data faster.

GRIDtoday: Specifically, what SAS product (or products) are involved in Grid computing?

NGUYEN: SAS helps to enable Grid computing with SAS/CONNECT's MP CONNECT functionality, online at www.sa s.com/technologies/dw/etl/connect/index.html. MP CONNECT allows users to segment a job workload into independent units of work to be processed in parallel across any number of computers within a network. In our newest software release, SAS 9 (available in a phased release starting Oct. 31), MP CONNECT's functionality will be extended to support piping, which allows dependent SAS processes to overlap their execution and eliminate the need to write intermediate results to disk. In other words, SAS/CONNECT automatically segments a large job into components that are easily consumable by the resources within the network.

GRIDtoday: Last question, and it's a sensitive one: If Grid computing is supposed to provide "supercomputer" performance to anyone who wants it on demand, how will we be able to keep terrorists from using it?

NGUYEN: This is a topic that sparks a lot of interest as it is on many people's minds -- which is typical with any new and emerging technology that comes to fruition.

I don't have all the answers, but I do know that the Globus Project has a toolkit that includes security measures that could potentially help control terrorist involvement. Many of the Global Grid Forum members are currently, or soon will be, involved in this organization. Part of the toolkit includes a component called the Grid Security Infrastructure, which provides secure authentication and communication over a Grid network. In addition, it supports security across organizational boundaries and single sign-on for users of a Grid, including delegation of credentials for computations that involve multiple resources. Smartcard support also helps to increase credential security. I'm sure that as time goes on, additional security measures will be developed and implemented to create a secure Grid environment for companies.

With the promise that Grid computing is already delivering on -- lower costs and boosted efficiencies of computing networks -- these and other issues will be addressed so that more organizations can safely and securely leverage Grid computing.

For more information on SAS, please visit the company's Web site at www.sas.com . The Global Grid Forum and the Globus Project can also be visited online at www.ggf.com, and www.globus.com , respectively.