Internet2 Building the Next Backbone

By VERONICA SMITH

Internet2 has been publicized in different lights, which means there is still lots of confusion about the project's internal goals and its ultimate purpose. Promising to provide access and transfer rates up to 1,000 times faster than tho se available today, the project is deserving of attention, but don't expect to be cruising at such high speeds just yet.

Currently confined to academia, government research centers and a handful of nonprofit organizations, the new network is separate from the mainstream Internet. As far as access goes, however, Internet2 is not considered a "separate" network. The most significant distinction that can made at this point is that Internet2 will not replace the current Internet. The hope is that the developments and discoveries that emerge from continuing work on the Internet2 project will eventually find their way into all kinds of networks, including the Internet.

According to Greg Wood, communications director for the Internet2 project, "Internet2 is not a separate network. Internet2 participants are not building new networks outside of their campuses. They are upgrading their existing campus networks and establishing connections to GigaPOPs by acquiring network services from existing providers."

GigaPOPs are hubs with high-speed connections-"giga" for billions of bits per second, and "POP" for points of presence, or the connection where the local carrier hooks up with a long distance circuit.

According to the Internet2 project working group, "the goal of Internet2 is to facilitate and coordinate the development, deployment, operation and technology transfer of advanced, network-based applications and network services to further U.S. leadership in research and higher education, and accelerate the availability of new services and applications on the Internet."

Internet2 uses existing networks, such as the National Science Foundation's very-high-speed Backbone Network Service (vBNS), and will eventually employ other high-speed networks to connect members to each other and the outside world. Part of the mission is to make sure that all future developments in hardware and software are based on open standards and that the technology is readily available to commercial network users and Internet service providers. "Again, it's better to not think of Internet2 as a completely separate network, but rather a technology development effort," says Wood. "A major goal of Internet2 is to ensure that the technology developments find their way into the broader networking community."

Offloading academic and research-oriented traffic from the commercial Internet will speed data transmissions for users of both the Internet and Internet2. But in addition to providing a much faster network infrastructure, Internet2 applications will take advantage of a new set of networking tools, those that will arise out of the experimental work of Internet2 itself. One of the priority tools to come out of Internet2 research will be quality-of-service guarantees, which do not exist on the current Internet. Right now, all data that flows on the Internet is given the same priority as it is passed along the network from one computer to another. Quality of service will set priority levels for information that is sent out.

The development of leading-edge applications, such as tele-immersion, digital libraries and virtual laboratories, is another major goal of the Internet2 project, and the supporting network technology will be developed as the new applications arrive. The Internet2 project's experimental approach allows for the development of innovative network features and improved functionality that can be easily transported to the commercial Internet, which is why there has been significant contribution and funding made by large corporations in the industry, including Cisco, IBM and MCI.

The Role Of vBNS

Among these contributions, Internet2 chiefly relies on the NSF's and MCI's vBNS. So far, more than 50 institutions have received competitively awarded vBNS grants under the NSF's High Performance Connections program.

Launched in April 1995, the vBNS was designed as a high-bandwidth network for research applications. It's a nationwide network that operates at a speed of 622 Mbps (OC-12) using MCI's advanc ed switching and fiber optic transmission technologies. MCI offers this example: At speeds of 622 Mbps, 322 copies of a 300-page book can be sent every seven seconds.

vBNS can be considered the pulse of Internet2. Under a five-year project with MCI, which contributed $50 million, the service links six NSF supercomputer centers that support "gigabit test beds" for advanced networking research and development.

Some of the work being done includes ATM/Sonet, the interfacing of ATM to the High Performance Parallel Interface switches and all-optical networking. Each test bed addresses applications that require gigabit-speed networks.

The vBNS provides high-speed interconnection among supercomputer centers, research and education institutions, other research networks and NSF-specified network access points. The service offers high-speed data connectivity on an institutional level and gives remote users access to high-performance supercomputer centers. The network further supports distributed processing , which allows users access to resources at two or more supercomputer centers and creates what the group calls a supercomputer "meta-center." The vBNS also supports Internet video teleconferencing and workgroup tools to aid in collaborative research.

In addition to offering a high-performance network service, the vBNS cooperative agreement will generate a new breed of hardware, protocols and transmission technologies, and integrate them into the network as they become available. To support the NSF's goal of the continued development of high-performance networking, the agreement also encourages efforts to develop and deploy advanced routing mechanisms, protocol enhancements and performance measurement techniques.

Next-Generation Internet Applications

So far, more than 100 universities and research institutions are onboard with the project, while 53 Internet2 universities have won High Performance Connections awards from the NSF. At least 20 of those are connected to the NSF/MCI vBNS.

The project will be conducted in phases over the next three to five years, in line as much as possible with the Clinton administration's Next-Generation Internet (NGI) initiative. In the first phase, end-to-end broadband network services will be established for the participants. At the same time, teams of university faculty, researchers, technical staff and industry experts will begin designing the advanced applications. By early 1999, there should be numerous beta versions of applications in use on Internet2.

The main efforts will exploit the broadband networks' media integration, interactivity and real-time collaboration. The group is developing new applications to further enhance researcher collaboration for conducting experiments; an enhanced delivery system for services such as health care and environmental monitoring using the "virtual proximity" advanced communications infrastructure; middleware and development tools to support the advanced applications; better and more affordable methods of technolog y development, deployment and operation of an affordable communications infrastructure; and working standards and common practices among the participating institutions to ensure quality of service and interoperability.

The applications heading the list for Internet2 are collaborative environments, digital libraries, tele-immersion, tele-medicine and distance-independent instruction. Projects under development with Internet2 include digital multimedia libraries accessible in virtual reality, enhanced collaborative workplace communities with live digital video feeds, videoconferencing, collaborative computer-integrated manufacturing, weather forecasting and military-troop-movement monitoring.

The Internet2 group is planning an entirely new set of demonstrations for April 14-16 that will feature the new set of applications. See www.internet2.edu/oct97/html/demos.html for some examples of applications being developed.

Internet2 Architect ure

The architecture for Internet2 applications follows the model in which the client side relies on component for construction, and middleware services that use operating system functionality will communicate over the network to application and network services. The server-side employs an "n-tier" model in which multiple servers may be applied to a single application. The Internet2 working group strongly states that these concepts are just guidelines, however, and that they function only as a starting point for the work in progress.

The architectural model for Internet2 addresses several major areas. The first main focus is to make components, such as objects, modules and other interoperable software parts "integrable," which means they can easily work together or separately, depending on the task at hand. There also will be the introduction of new component classes for authentication, authorization, and validation services, directory services, presentation services and simulation parts.

Rese archers are going even further with components, though, because real-time collaboration has become a requirement for them. This means new services allowing for the run-time access brokerage to components, objects, and information resources must be added. These component brokerage services are necessary for "run-time binding" of all the various parts, or modules, to be integrated into "operable" applications.

Support for compound documents also will be added to Internet2. The support will be built into new frameworks, as well as into the Web. Under this infrastructure, content can be more flexibly placed, objects can be embedded, and passive and active information elements will be accommodated. Services for information resource metadata will be added to act as application components and a data repository so that all researchers from each GigaPOP will be able to access the same information. To facilitate this new functionality, the Internet2 group will continually work to standardize APIs for various forms of application support and service.

On the network service level, capabilities will be integrated so that application components can negotiate for services, such as increased bandwidth, quality-of-service guarantees or latency levels.

To accommodate the networkwide testing, benchmarking and other forms of usage measurement, special performance metric applications will be created and deployed.