Proposal for an Internet Engineering RepositoryTable of Contents
- Introduction and Background
- Proposed Research
- Project Management
- Sample Initial Topic Areas
- Evaluation and Sustainability
- Statement of Work, Cost Estimates, and Timelines
- Prospective Contributors and Support
- Selected Bibliography
This proposal requests incremental funding for the National Laboratory for Applied Network Research (NLANR) project to build a living, on-line repository for network engineering instruction materials.
Continual exponential growth in traffic and bandwidth usage is putting severe strain on the Internet, and the gap between the immediate needs of the Internet infrastructure and current classroom content is growing. New computer science and engineering solutions are needed, but producing the necessary skills and talent is hampered by the small size of the current pool of knowledgeable network engineering practitioners. They, in turn, are hampered by the prevailing hand-to-mouth style of knowledge transmission.
Under the auspices of NLANR, UCSD proposes to coordinate a Web-based repository for a network engineering curriculum materials. Our resources are the current instructors and Internet vendors such as CISCO who have a wealth of training materials related to their hardware and software, and our objective is to draw together the myriad topics in a continually updated resource for both teachers and students. A central trove of engineering materials can be instrumental in creating a stable network infrastructure whose reliability and throughput at various levels will be secured by the activities of engineers, users, and service providers alike.
Introduction and Background
Ensuring the most intelligent evolution of the dynamic Internet infrastructure is tied to getting both broad and deep study of that challenging environment into the classroom. Several times in the past two years, and more and more frequently, components of the network have collapsed under the weight of the traffic. As ever more complex and compute-intensive applications involving audio, video, and 3-D visualization are loaded onto a system already severely strained, the prospect of catastrophic breakdowns looms.
What is being taught in networking courses at most American universities is not out-of-date by most standards. It's "hot off the press." The information in textbooks is only three or four years old, and the ancillary readings are even newer.
But ask which students are familiar with interactions among the communications protocol layers, the effects of cell switching on current Internet environments, accommodating diverse service qualities, hierarchically coded multimedia data streams, or reservation of network bandwidth, and you will find that these topics, only a year or so old in the Internet engineering community, are not treated at all or barely mentioned. The same question might be asked of Internet engineers already in the field who are faced with continually changing technologies and technical advances. While such a gap can be tolerated in most sciences, it must be erased as quickly as possible in this field, because leaving the network's foundations at risk is a costly proposition.
The community of up-to-date individuals is very small, and teaching is typically not one of their priorities. They are holding their fingers in the dike, substituting tirelessness for the infusion of engineering talent needed for real solutions. Most of them belong to the Internet Engineering Task Force (IETF). A handful of the several hundred IETF participants are also instructors in engineering and computer science departments at major universities. In the classroom, they are waging separate and isolated battles to include more timely topics in their curricula. We believe it serves the long-term interests of advanced networking research to improve this situation by applying the object of study itself to multiply the efforts of this group.
UCSD proposes to sponsor NLANR coordination and operation of a Web-based `living repository' for Internet engineering. This project is not attempting to define the beginning and end of an Internet engineering curriculum, but rather use the Web as a collaborative resource to which instructors and practitioners contribute their knowledge and from which this knowledge will be distributed to the wider Internet engineering community.
Such a repository will contain unpublished classroom materials, published research papers, problem sets and solutions, real network statistics, software tools for Internet engineering analysis and for distance learning, and possibly even real-time conferencing on Internet breakdowns as they happen. Unlike a published book, a Web repository offers the advantages of continual expansion, revision, reorganization, and revitalization.
Instructors will be able to borrow from one another in producing classroom materials for certain topic and focus their energies on emerging topics or gaps in the collective coverage. In an arena as rapidly changing and dynamic as the Internet, this repository will take "hot off the press" to an entirely new level. By contributing material for use by groups of colleagues nationwide, the current teachers can greatly increase their effectiveness. We are requesting funding for a three-year program, including a full-time site coordinator as well as servers and storage media on which to maintain the materials at one main site and two or three mirror sites.
As a multi-institutional effort to support research on the advanced, high-speed wide-area network connecting supercomputer resources at NSF-sponsored sites, NLANR is a logical host for such a site, and the repository will leverage both NLANR expertise and resources at NLANR sites. A recognized, objective focal point for network infrastructure research, NLANR members are actively seeking solutions to the looming and philosophically murky puzzles of the Internet. NLANR unites efforts in measuring the unknowns about capacity, traffic, and future planning. While the growing group of Internet service providers has interests in such questions, they typically have neither the time nor the expertise required to pursue such dedicated engineering investigations. Owing to widespread misunderstandings of the fundamentals, indeed, some service providers are actually a source of instability and strain on the net, however they might wish to be otherwise.
NLANR researchers at UCSD will not only provide guidance on the organization and relevance of materials in the repository, but also be a fertile source of real-world, cutting-edge contributions. Current NLANR projects include, for example, 3-D visualization methods of network architecture, Internet statistics gathering, vBNS applications, global Web caching, the proposed Cooperative Association for Internet Data Analysis (CAIDA) to promote industry cooperation in managing the Internet, and the IPNmoo tool for real-time interaction during diagnosis and repair of network problems.
By virtue of their recognized expertise, nationwide collaborations, and locations at leading research universities, NLANR researchers also have contacts and recognition throughout the Internet engineering community. They will be able to direct the coordinator to likely contributors in virtually every area of Internet expertise.
As a distributed laboratory, NLANR also has the advantage of its researchers being located at five national laboratories for high performance computing and communications. These sites, such as the San Diego Supercomputer Center (UCSD's SDSC), are involved in high-speed networking projects and are also charged with making high performance computing resources available to the national research community. The repository coordinator will be able to leverage such resources, such as data archives at UCSD/SDSC, as they become necessary for the project. UCSD/SDSC also has researchers experienced in similar repository projects for the Virtual Reality Modeling Language (http://www.sdsc.edu/vrml/) and for the protein kinase family of biological enzymes (http://www.sdsc.edu/kinases/).
These five NLANR sites have further connections to the Internet engineering community by being located at universities with leading departments of computer science and engineering. Instructors from these departments will be contacted to contribute their knowledge to the repository. Rather than being isolated from the classroom, NLANR offers a potent combination of a distributed laboratory addressing real-world Internet development, national laboratories in high performance computing and communication, and university campuses to nurture the next generation of Internet engineers.
Funding of $379,409 is requested for a three-year effort, including regular evaluation of the site's impact. In addition to the appropriate equipment (server) and disk space, UCSD proposes to support a coordinator, or coordinators, to maintain the site and disseminate information about it, seek out new participating teachers, evaluate its success, archive curriculum materials, and assist contributors, including those who are also textbook authors (and their publishing agents), with copyright and licensing issues.
Repository Coordinator. To oversee the repository, UCSD will hire one full-time equivalent (FTE) as coordinator. The coordinator duties may be held by one person, or several (graduate students, for example). The coordinator's duties will be to gather content, maintain valid URLs, identify gaps and request materials to fill them, verify suitability of contributions and security of downloadable software, organize contributions into logical topic areas, and maintain the infrastructure to sustain a public, net-wide discussion of research choices and modes.
The role will be "coordinator" deliberately distinct from "editor." The project coordinator will not be able to "edit" repository materials or create a finished product in the traditional publishing sense. As the repository reaches critical mass, items may be superseded, software may be updated, new topics may be added and old ones archived. The coordinator will be charged with maintaining a coherent organization and superficially evaluating a contribution's "safety" (as with software, for example) or verifying a contributor's credentials. By contributing and using the repository materials, users will reach consensus on the most useful items and provide feedback to the coordinator on those with inaccuracies.
We also note that the Internet engineering topics are the subject of considerable heated exchange on mailing lists and at IETF working group meetings. One of the responsibilities of the coordinator would be to distill discussions and minutes from the meetings into digestible updates to the respective Web pages. This service would be valuable not only to instructors but to members of the Internet community themselves seeking a "Readers Digest" version of those proceedings.
NLANR PI. The NLANR principal investigator(s) will supervise the coordinator and provide guidance where necessary. They will also perform a vital role by staying abreast of current research, contributing relevant materials from their research, and informing colleagues of the repository resources.
Repository Site at NLANR. One advantage of NLANR is that the physical location of the coordinator or coordinators and the server hardware can be at a number of sites. NLANR researchers and site coordinator can leverage resources (data storage, computing cycles, administrative support) from any one of several NSF-funded facilities as particular needs arise.
Sample Initial Topic Areas
To illustrate what we mean by a "living repository," we have assembled an initial list of topics for which materials could be invited or solicited (see Attachment 1). The proposed budget includes funding for honoraria to provide incentive to contributors for key areas. A sample bibliography covering several of the topics is included; one objective would be to have links to sites where such articles may already be kept. Failing that, the coordinator would try to get suitable permissions to make readings available in the repository itself. When useful, an attempt will be made to produce lectures and other material in more vivid form, as animations with voice-over or as informal video clips of the lecturers or instructors themselves.
Many of the topics below concern the intimate relations between network engineering and specific applications carried by the network. In an infinitely redundant system, the nature of the bits transferred would not matter. But the current network may stand or fall on common understandings of what the traffic carried actually represents in computational terms. Moreover, the interactions required to solve problems related to the networking of specific kinds of applications may themselves be the wellsprings of future advances in networking.
Additionally, completely new disciplines are arising in the legal and policy arenas, which network engineers will be increasingly unable to ignore, since they will be confronted every day with dilemmas in the areas of privacy, security, copyright, export controls, intellectual property, and license violations, among others. The repository will incorporate these social aspects of networking; they are too essential to ignore.
A prototype repository, organized according to the broad subject areas below, has been implemented at http://iec.nlanr.net/
The sample repository topics exemplify the breadth of subject matter appropriate for the repository. Once implemented, the repository will grow broader and certainly deeper based on materials from contributors and developments on the Internet. Contributors would be encouraged to identify gaps and provide supporting materials.
Evaluation and Sustainability
The coordinator will take several approaches to evaluating the repository, located on a Sun Sparc Server 1000E at the UCSD's SDSC. Throughout the three-year project, unsolicited feedback from contributors, browsers, and users will suggest minor (or less often, major) additions or reorganizations of the repository.
In addition to counting "hits" to the Web site, contributors and registered users would occasionally be surveyed on the utility of the repository. By asking users to register, the project will have a better estimate of the most useful and popular aspects of the repository. Evaluation will also confirm that the site has been adopted by the university and instructional community and is reaching, or has reached, the critical mass required to make it self-sustaining.
The true measure of the project's success will be the willingness of the Internet community to support the repository beyond the three-year proposal period, and the coordinator will follow two possible avenues toward sustainability. On one hand, a university department, or several universities, may adopt the site as a vital component of a newly developed Network Engineering Department and continue to provide the service to the networking community.
On the other hand, the repository will also serve to give commercial publishers a sense of the market for materials that can be accessed well ahead of the current timetable for bringing a textbook to market. Publishing houses may determine that such repositories may be economically viable as an on-line publishing methods, not only on Internet subjects, but also in other disciplines.
Statement of Work, Cost Estimates, and Timelines
If this proposal is approved, NLANR will have a machine and storage for the site in place within six weeks to two months after funding. The job of coordinator for the site will be filled at the same time.
UCSD estimates the price to perform this activity will total $379,409 over a three-year period. Standard NSF 1030 budget forms are attached for the total proposal project budget as well as for the individual years. Supporting cost breakdowns are provided in Table 1.
The coordinator's first task will be to recruit contributors for the various sections. NLANR contacts will prove useful for beginning this process of collecting course materials, research data, tools, problem sets, and sites on Internet engineering topics.
Some of these authors are likely under contract for books on these topics, since they are of such broad interest. Many publishers have found that it aids in selling advance copies of a book for a selected chapter to be out on the net. The coordinator would make suitable arrangements with authors and any other holders of copyrights to materials to be posted to the site.
Similarly, the coordinator would make license or other agreements concerning software that can be downloaded or run from the site for problem sets or other classroom projects, and the coordinator could install grading facilities if site users desire them.
The coordinator would advertise the existence of the site to university instructors of undergraduate and graduate courses in networking, network engineering, and computer science.
The coordinator will work with site users and NLANR to devise an evaluation procedure to apply periodically to the site and its contents, which we expect to change over time, with flexible procedures for users to update/add materials. As topics become dated, perhaps by being adequately represented in adopted texts -- the repository will, of course, include a bibliography of such texts -- we will replace them with more timely topics.
The coordinator will also cooperate with motivated faculty in university departments to develop more formalized curricular units based on materials in the repository. With the networking research community, the coordinator will work to expand the site's role as a "window" into ongoing research efforts, such as vBNS, CAIRN, and others, by encouraging, for example, on-line discussions and debates.
A second-year objective will also be to install one or more mirror sites. One potential contributor has offered to oversee such a site in Great Britain (see Attachment 2).
Evaluation procedures will have been in place long enough to enable the coordinate to conduct a preliminary evaluation of the site's impact. On the basis of that evaluation, we can make changes and organize the lessons learned into a presentation that we will put on the site itself.
The close national coordination in curriculum development that the site will make possible may turn out to be a model for curriculum development in other fields as well, and thus we intend to make the results of the experience available as widely as possible. Travel money would be used during the third year to facilitate wide dispersal of the message via attendance and displays at meetings and conferences.
During the third year, the coordinator will also pursue contacts with university departments and publishers to take over the site and sustain it either as a free educational service or as a commercial venture (or both).
Prospective Contributors and Support
A draft of this proposal was recently made available on the Web and circulated to experts in universities and industry whose contributions to the literature suggest that they might be able to provide unique curricular materials. We have already received several enthusiastic responses via e-mail (see Attachment 2) and are thus encouraged that there is community support for this idea.
An example of the ancillary readings the repository would make available to students and faculty is a selected bibliography taken from the syllabus of a course taught at UCLA by Professor Lixia Zhang (see Attachment 3). Many were originally published in proceedings volumes whose distribution across libraries is spotty, so availability online is key.