An Applications Strategy for the NGI Ian Foster and Rick Stevens Argonne National Laboratory {foster,stevens}@mcs.anl.gov Need for Aggressive Applications The Next Generation Internet initiative is about change and creativity as much as technology. The NGI is motivated in part by the desire to keep the wonderful process going that has produced today''s Internet and to continue to drive and nurture that process into the next millennium. The early Internet was driven largely by technology. The arrival of the next- generation Internet will be determined not by the networking technology being used but by milestones in the way we think about and use the network. (For example, no one remembers when ATM backbones went into production, but everyone remembers when the Web became popular.) This change in the perception of users will occur when the applications we routinely use are different from those we use today. We believe that this transition will be determined to some extent by network technology (enabling some basic new capabilities) but to a larger extent by a breakthrough in how an application uses this technology. However, unlike microprocessor performance or even networking bandwidth, it is nearly impossible to predict the type of applications that will dominate the use of the Internet in the early decades of the 21st century. All that we can reasonably state is that they are likely to be significantly different from what we routinely use today. Need for Conceptual Revolutions We believe that while it is critically important to invest in networking infrastructure and middleware for NGI, it is perhaps even more important to create an aggressive applications program that will encourage and reward innovative applications concepts and demonstrations. The goal is to encourage networking and software technologists to strongly engage and support the applications community to produce creative new ways of using emerging network technology. Only by encouraging innovative uses of the network will we see the emergence of truly new application concepts that will justify sustained national investment. Teleimmersion and Shared Virtual Reality Teleimmersion provides the most fertile ground for the development of new applications concepts. Teleimmersion systems combine real-time audio, video, tracking and sensor data, simulation data, and immersive images. These systems are likely to yield entirely new types of applications that incorporate the best features of desktop environments, while overcoming the limitations of such environments. A key challenge for NGI is to provide to the end user or programmer fine- grained control of network latency, bandwidth, and jitter (variance in delay). Improvements developed for teleimmersion environments will immediately benefit desktop applications. Shared virtual reality spaces will challenge our notion of collaboration by enabling the creation of virtual worlds based on both real and imaginary entities and environments, including environments where people interact with and collaborate with computer simulations or agents. Shared VR spaces provide will also set challenging performance goals for networking development and systems software providers. Real-Time Remote Instrument Control Experimental science has been rapidly moving toward computer-controlled instruments since the early seventies. Today almost all major experimental facilities are controlled by computers. The next step is to enable remote use of experimental facilities by distant users (or operators). By developing the capability to routinely use sophisticated and expensive instruments remotely, we can improve the quality of science produced (by enabling remote investigators to be more involved in the experiments) and can improve utilization of facilities by reducing the time needed to become familiar with complex devices by supporting remote training. Remote instrument control requires secure network access and reliable authentication of users. Via NGI, schools and community colleges can begin to participate in large-scale experimental science, from radio astronomy to molecular biology. Collaboratories for Technology Development One benefit of the NGI will be the ability to build networked environments to support widely dispersed groups developing new products or technologies. An example might be the worldwide community engaged in developing nanotechnology. Since this community is geographically dispersed and heterogeneous, it needs access to remote databases, large-scale computers, modeling environments, and experimental systems, as well as to each other. These diverse resources need to be accessible via desktop systems, mobile devices and large-scale immersive displays. The NGI is a critical resource to support the emergence of these networked communities. Education and Research Media Libraries One popular vision for education environments of the future is the elimination of barriers to access. This may mean, for example, the ability to participate in live courses via high-bandwidth virtual reality-type interfaces or simply to have access to recorded materials in all fields of study, independent of one's location, educational level, or professional status. By enabling network technology needed to easily record audio, video, and data to network-based repositories (i.e., directly to remote network servers rather than to tape), in a few years it should be possible to capture for playback any publicly available audio and video session on the Internet or to make entire libraries of recorded material available to remote users. An important goal for NGI should be to enable the remote ability to create, manipulate, and access multimedia datastreams as easily as we manipulate text and still images today. Creating an NGI Applications Forum Creating the right environment for NGI applications to grow and flourish requires work. Applications development activities need to be strongly and explicitly encouraged (even if not explicitly funded by NGI). We believe that the leading federal agencies for NGI (DOD, DOE, NSF, NASA) should create or encourage the creation of an "Applications Forum" that will help provide mechanisms for applications developers to interact with and to help drive the networking technology development supported by NGI. The forum does not have to be complicated or expensive. It needs simply to provide an opportunity for interactions that otherwise may not happen. Supporting Advanced Applications Development We also believe that advanced applications development needs to be supported by the same agencies that are supporting the NGI, though probably from other funds. Applications development and demonstrations should be an integral part of any NGI testbed or networking upgrade plan. Events such as the I-WAY at SC'95 demonstrated that significant applications development can occur with minimum direct financial support if the applications community is provided with a mechanism and opportunity to participate in a national-scale event. National Events to Drive Technology and Applications Demonstrations Accordingly, we believe that the NGI should sponsor an annual national event that demonstrates the advanced in networking technology, software environments, and applications. This event could be associated with a conference or other activity, but it should be high profile and applications focused. Such national events provide a strong deadline-based driver for technologists to aim for. National events also provide the publicity and exposure that applications groups need to continue their investment in leading-edge demonstrations.