Susan Estrada Acting Director, CENIC susan@aldea.com 760-929-0580 (phone and fax) CENIC 2380 Camino Vida Roble, Suite A Carlsbad, CA 92009 Seeking to support growth of the Internet, Susan founded Aldea Communications, Inc. in July 1993. Aldea's focus is to advise companies and universities on strategic telecomunications strategies. Its client list includes the University of California, Hughes, AT&T InterNIC, Network Solutions, Cisco Systems, AT&T Jens, Pacific Bell, and Bell South. Aldea's current focus is the next generation Internet and higher education. Susan, as acting Executive Director, is providing administrative and technical support services to the Consortium for Education Network Initiatives in California (CENIC), a next-generation Internet project for research universities in California. In addition, under Susan's guidance, Aldea has created the InterNIC Academic Guide to the Internet under AT&T's InterNIC Directory and Database Services NSF cooperative agreement. Giving practical advice on getting the best Internet service, Susan wrote Connecting to the Internet, An O'Reilly Buyer's Guide, in August 1993. The book made Barnes and Noble's best seller list. Always forward thinking, Susan founded CERFnet, an Internet service provider, in 1988. During her 5 year tenure as the CERFnet executive director, she was instrumental in CERFnet's user growth from 25 university members to hundreds of corporate members and thousands of individual users. As a parent of two young children, Susan is strongly committed to enhancing educational programs using communications technology. Among other activities, Susan played an instrumental role by co-founding the Global Schoolhouse and organizing the California branch of Tech Corps. Susan is an elected Trustee of the Internet Society, a founder of the Commercial Internet Exchange (CIX), a former area director for the Internet Engineering Software Group (IESG) and the Internet Engineering Task Force (IETF). She currently is an appointed member of Pacific Telesis's Telecommunications Consumer Advisory Panel (TCAP) and the U.S. Federal Networking Council's Advisory Committee (FNCAC). An innovator and entrepreneur, Susan Estrada's experience includes almost 20 years association with telecommunications and network management for companies such as AT&T and Pacific Bell. -------- The Need for NGI The CENIC Perspective Susan Estrada Acting Director, CENIC The Consortium for Education Network Initiatives in California, CENIC, is working with the key universities in California to develop a high-speed state-wide network that focuses on the precompetitive Internet environment - 12 to 24 months ahead of the commodity Internet. One of the questions that is oft repeated is "Why does the Internet industry need projects like CENIC? All of the problems it is planning to solve will be solved by the commodity Internet Service Providers, ISPs." This paper will present reasons why this is not a valid statement and the value and necessity for projects like CENIC in advancing Internet architecture. Because of the nature of an organization like CENIC -- diverse participants with wholly different goals -- one can think of a CENIC network as a backbone connecting a variety of campuses acting as micro-ISPs. The participants in CENIC include private universities such as Stanford, USC, and Caltech as well as California's two public university systems, the University of California and the California State University. Each of these institutions has diverse needs. Some of the campuses focus on research. Many of the campuses instruct large student bodies. Both of the public systems are budget restrained but have the need to network their campuses throughout the state. Each institution has a different budget, different networking skill levels, different institutional requirements for delivering networking to faculty, staff, and students, different campus policies, and the obligation to obtain the highest quality networking for the least cost. Because of the diverse nature of the CENIC participants, CENIC believes that many problems that effect typical Internet users could be modeled and solved effectively in the CENIC environment. Because of the nature of the current commodity Internet -- heavy competition, low margins, extreme growth -- these problems are unlikely to be solved because they require significant ISP cooperation. In many cases, this cooperation would result in the lessening of the uniqueness of an ISPs perceived value added service. This makes such cooperation highly unlikely without outside pressure or proven, customer acceptable solutions. There are some clear areas where a collaborative like CENIC can play an essential role in developing technologies that foster the implementation of the Next Generation Internet (NGI). These are the same areas where ISPs fear to tread because of the nature of the commodity Internet. "A friend and colleague of mine often refers to what he calls the dancing bear phase of new technology. An admiring crowd surrounds this wondrous performer, saying, "Oooh," and "Ahhh" and "Wouldja look at that? It's a by- God dancing bear!" It is only later, says my friend, that anyone bothers to ask how well the bear can dance, or whether dancing bears are somehow useful." -- Thomas R. Bruce The Internet was a dancing bear for many years. The bear danced well enough to meet the early expectations, such as e-mail, file transfer, the Web and even some primitive video conferencing. Now, however, higher education is substantially increasing its expectations for educational networking. Some uses, such as video conferencing in support of instruction, require more than Internet-style "best effort" network. If faculty are to be able to rely in the Internet for this class of applications, they must be provided with an advanced services Internet that supports differentiated modes of transport. In many instances, higher education will decide that as a matter of policy certain types of traffic must not be adversely affected by certain other classes of traffic. For example, some institutions may adopt a policy that real-time multicast video must be given priority over IP unicast traffic. Students downloading large data files will simply have to wait longer so that the quality of instructional video is not adversely impacted. QoS assurances are intended to assert that the network will provide services which conform to any of several explicit, quantifiable service parameters. This is in sharp contrast to today's Internet which provides only "best-effort" service. These service parameters may be chosen by the network operator, the campus network managers, or the end users. CENIC members will be working on middleware and applications libraries that will extend QoS capabilities to desktops in support of the full range of potential applications. CENIC will address the following QoS concepts: Path guarantees; Assessment techniques including service tracking and quantification of QoS data; Charge back schemes to evaluate the cost/benefit of QoS; Ultimately, however, bandwidth is a finite resource, even in a network like the one CENIC is planning. That's where policy comes in. A group of people, not software, needs to determine whether video of a scheduled class ought to be preempted to provide medical imaging for an emergency procedure. The CENIC Policy Development Working Group will tackle these issues. One of its first tasks is to explore how "human readable" policy statements translate into network functionality. For example, how does "This class was scheduled a year ago, and so must have priority unless the Dean approves otherwise," translate into a router or switch configuration. One of the key aspects of QoS is understanding what the network is doing. Adequate data collection and analysis is fundamental to understanding network performance. CENIC's network will be equipped with hardware to facilitate data collection for tracking service quality, distribution, predictability, and developing performance profiles. Capturing this information will allow CENIC member engineers to determine an initial baseline of performance and quantitatively measure the effects on performance as new traffic emerges and software begins to take advantage of differentiated services. Additionally, this information will be shared widely with the Internet community -- unlike the closely held network performance data of most commodity ISPs. CENIC's network is going to be expensive. Institutions will likely make substantially different uses of the network. It is recognized that a simple "divide by n" strategy for allocating cost will not be adequate when some uses of the network may require large blocks of dedicated bandwidth while other uses will simply use whatever bandwidth is left over. CENIC's Cost Allocation Working Group will define these issues and suggest methodologies for implementing charge-back strategies, particularly for applications which use QoS. This starts to identify issues that address settlements between ISPs and effective, acceptable payment schemes for consumers. Applying Quality of Service (QoS) techniques holds the promise of allowing us to make next generation Internet technology useful and affordable to the widest audience. The current single service Internet requires that everyone pay for the same class of service and argues for an "average level of service" for everyone. A researcher may require, and can be able to pay for, better service but there is no way to sell it today. Differentiated services will allow users to choose the most cost effective level of service to suit their needs. Not only are QoS techniques essential to the academic community now, but they will drive the success or failure of the commodity Internet of the future. References: CENIC http://www.aldea.com/cenic