Mission

This document represents an early attempt to articulate the goals, scope and direction of the End-to-End Performance Improvement Environment or “E2Epie.”

This formulation of E2Epie builds upon the foundation established by the Internet2 End-To-End Performance Initiative’s design document (February 28, 2001), which defined the need for a “knowledgebase” as a critical component of a broader Performance Evaluation and Review Framework (PERF).

The mission and vision of E2Epie have been extended and refined through a series of (14) interviews with members of the Initiative’s design team. Interview notes are included as Appendix I. Please understand that the notes are paraphrased rather than verbatim transcripts. Also, names and affiliations of interviewees are not included as we didn’t solicit permission to distribute publicly.

Our definition of E2Epie is very much a work in progress. We are distributing this document early in the process as a catalyst to provoke feedback and discussion from the broader community of Internet2 members and partners. We very much want to hear from you. What do you expect from E2Epie? What advice do you have? What should we avoid? Who should we talk to?

In addition to incorporating your feedback, our next steps include:

§ Conduct a second series of interviews with a broader group of people representing such diverse communities as faculty, researchers, campus technical staff, university CIOs and application developers.

§ Create a project plan, staffing model and schedule for E2Epie design and implementation over the next two years.

§ In parallel with long-term planning efforts, launch an initial web site by September 2001 that pulls together existing content and services related to End-to-End Performance, and that invites broad contributions to the evolving definition of E2Epie.

Please direct your questions and suggestions to:

Cheryl Munn-Fremon

Director, Internet2 End-to-End Performance Initiative

cmfremon@internet2.edu, phone, address

Mission

The End-to-End Performance Improvement Environment or “E2Epie” is a vital component of the Initiative’s broader goal to enable the end-user to obtain optimal performance from the current network infrastructure.

As recent experience within the Internet2 community has demonstrated, despite high-bandwidth connections, campus researchers, faculty, students, and staff often don’t experience the performance they expect from their networks. Fat pipes are not enough!

The Initiative was founded with the understanding that these performance problems cannot be solved by technology alone and that they can’t be successfully addressed by any one organization operating in isolation.

Rather, we need to find ways to facilitate collaboration and knowledge management across hundreds of campuses and labs, helping people to help each other to manage expectations, measure performance, identify problems, share tools and information and experiences, and define real solutions within their networked computing environments.

Vision

Stated simplistically, E2Epie is a web site hosted within the Internet2 web site http://www.internet2.edu. But thinking about E2Epie in this one-dimensional manner is self-limiting. E2Epie is a directory of network performance applications and tools, people, organizations, and projects. E2Epie is a digital library of technical reports, articles, books, presentations, web sites, white papers and training materials. E2Epie is a knowledge base of case studies, best practices and technical support documentation. E2Epie is virtual Network Operations Center (NOC) that accepts, analyzes, escalates, redirects and solves problems. E2Epie is an online community of people with diverse disciplinary and technical backgrounds sharing experience and expertise. E2Epie is a research laboratory, providing a rich context for studying the technological, economic, social and organizational opportunities and implications presented by the high-performance network environments of the future.

Each of these metaphors adds to our understanding of what E2Epie can be and do, but none of them can stand alone. We’ve chosen to define E2Epie as an environment that connects people to people, people to content, and problems to solutions. But we recognize the limitations of any definition. So, in the spirit of Yahoo and Google, let’s simply call it E2Epie or the PIE, and allow its evolving components and capabilities to shape our understanding over time.

Goals

Goals that emerged from the interviews include:

§ Enable geographically isolated people to find each other and communicate.

§ Help people to find and share information, experiences, applications and tools.

§ Set reasonable expectations of performance.

§ Coordinate support across campus, regional, gigaPoP and backbone networks.

§ Provide a framework for identifying and solving performance problems.

§ Close the “Guru Gap” by educating and empowering junior technical staff.

§ Reduce demand on the backbone NOC.

§ Enable better utilization of the Abilene backbone.

§ Leverage the potential of the high-performance network environment and community as a context for multi-disciplinary research.

Audiences

We think of the users of E2Epie as participating members of a community that spans diverse virtual and physical environments primarily within the Internet2 membership. Sometimes they will interact online. Sometimes they’ll meet face to face at conferences and seminars. Each member may wear many hats, performing different roles for different institutions at different times.

For these reasons, it’s dangerous to lock into a strict definition of audience segments and extrapolate towards assumption of individual needs. However, there is value in exploring a loose categorization of anticipated audiences.

Our discussions so far have uncovered four primary and four secondary audiences, as presented in the following table.

Table 1. Intended Audiences

Importance	Audience	Similar Roles & Titles	Notes
Primary	Network Engineers	NOC Staff	Easiest for Internet2 to reach. Early adopters. Both contributors and users. Operate at campus, regional, gigaPoP and backbone levels.
	Faculty Researchers	End Users	Self-identify within disciplinary communities (e.g., physics, astronomy) before E2E performance community. Experience with E2Epie may be mediated. Campus and independent labs.
	Application Developers	Researcher Support Staff, Programmer/Analysts	Major “mediated” connection to faculty researchers.
	Technical Support Staff	System Administrators, LAN Administrators, IT Liaisons, Help Desk Operators, Call Center Staff, User Support	Important audience that’s difficult to identify and reach. Mediate experience for faculty researchers. Operate at central campus and departmental levels.
Secondary	IT Executives	CIO, IT Administrators	Comparative statistical data, Return on Investment (ROI), case studies, best practices.
	Students & Faculty	End Users	Primarily mediated experience.
	IT Architects	Network Researchers, Data Mining Researchers	Operational network experiences inform next-generation designs.
	IT Vendors		Opportunity to learn from their customers and partners.

Principles & Guidelines

Based upon our research, experience and the advice offered during the initial series of interviews, we have developed the following set of principles and guidelines to shape the design and implementation of E2Epie.

§ Provide active leadership and facilitation. Experience has proven that significant effort is required to launch and sustain online communities. This is not so much a technology challenge as one of collaboration and knowledge management. We are committed to investing staff and resources to build a critical mass of content, contributors and users, so that E2Epie can become a vibrant community of interest and practice.

§ Build trust. E2Epie must be known as a trusted source of accurate information, exercising sound editorial judgment over resource selection and evaluation. High quality design, organization and indexing are also important factors.

§ Create an economy of recognition. Successful online communities operate as economies where reputation, recognition and attention are forms of currency. The E2Epie model will recognize and reward contributors, and will enable members to evaluate each others’ contributions.

§ Leverage existing resources. E2Epie will strive to be an “aggregator” that leverages, pulls together or points to existing resources and expertise, avoiding the unnecessary duplication of effort. This requires awareness of other initiatives and projects (e.g., directory of directories, Shibboleth).

§ Think big and start small. There is tremendous potential for E2Epie to become a powerful enabler of collaboration and knowledge management, making major contributions to the ongoing goal of end-to-end performance improvement. To fulfill that potential, it’s important to start small and design iteratively, so we can learn from our users and shape E2Epie accordingly.

Major Components

Throughout the discussions and interviews, E2Epie has been envisioned as both a resource (e.g., virtual library, online community) and an operational environment (e.g., performance measurement tools, knowledge base).

Our plan is to begin with E2Epie as a resource, focused on connecting people to people and people to content. This will allow us to provide real value in the short-term. In parallel, and working as a community, we will continue developing the requirements and framework for an operational environment that connects problems to solutions.

Ultimately, success of E2Epie will be highly dependent upon quality of execution and integration. Great ideas do not automatically lead to great successes. Neither do great components. The most useful and usable web sites are well conceived, well designed, well implemented and well maintained. The whole is more than the sum of its parts.

Given that caveat, the definition and prioritization of major components will play a critical role in shaping E2Epie. Discussions of E2Epie as a whole can be reminiscent of the fable The Blind Men and the Elephant. By focusing our attention on one component at a time, we can make sure we’re all talking about the same part of the elephant.

We’ll use the following 3 conceptual categories as an outline for exploring the mix of possible components:

§ People to People

o Directory of People & Organizations

o Communication & Collaboration Environment

§ People to Content

o Directory of Projects, Applications & Tools

o Case Study Database

o Digital Library

§ Problems to Solutions

o KnowledgeBase

o ExpertConnection

Directory of People & Organizations

A top priority early in the process will be the development of an online directory that connects people to people within the end-to-end performance community.

Demand for this resource is driven by:

§ The wide geographic distribution of Internet2 members in particular and the networking community in general.

§ The need for people at different institutions and in different roles to communicate with each other to identify and solve performance problems.

Several interviewees remarked that it’s really difficult for networking professionals on one campus to find their counterparts on other campuses. Communication today happens through informal, incomplete personal networks.

It will be important to model the complex relationships people have with their organizations. Within this community, people wear many hats through affiliations with multiple organizations. The mapping of these relationships will enable important types of queries. For example:

§ Show me all the organizations involved in this area of research.

§ Show me who’s doing work in this area at this organization.

Design guidelines for this directory will include:

§ Focus on Data Quality. The reality of any directory is that content is king. People want a resource that is accurate, current and comprehensive. Experience suggests people won’t reliably update their own data. We will commit staff to data entry and quality control, and we’ll also explore opportunities to leverage existing, distributed, authoritative sources of data (e.g., campus directories).

§ Provide Rich Metadata. Users’ ability to search, browse, and use the directory will be dependent on the selection of useful metadata fields (e.g., personal name, title, organizational affiliations, email, phone, fax, address, keywords, events attended, areas of expertise, home page URL, last updated).

§ Support Cross-Database Integration. There is great potential to support cross-linking between the directory of people and other areas of E2Epie such as discussion forums and the case study database. These connections between content, contributions and people will be critical to jump-starting an economy of recognition.

Communication & Collaboration Environment

A major goal of E2Epie is to facilitate communication and collaboration (one to one and many to many) among members of the end-to-end performance community. We are committed to identifying the best ways to accomplish this goal, independent of any particular technology or mode of interaction. We anticipate that preferred means of interaction will vary by audience, purpose and the number of participants.

The Communication & Collaboration Environment (CCE) will initially launch with a mix of the following tools and resources:

§ Email discussion lists (essential in building a committed group of regular participants)

§ Web-based discussion forums and archives (allows visitors to explore past discussions, make one-time contributions and evaluate whether they’d like to join a list).

§ Calendar of Working Groups’ Activities (e.g., telephone conference calls, face-to-face meetings, etc., etc.)

§ Instant messaging service

§ Guide to video-conferencing

We hope to use the CCE to provide an operational service and to conduct research into the strengths and weaknesses of the full spectrum of communication technologies and modes of interaction, adding new tools over time.

Early success factors will include:

§ Appropriate Interaction Mode(s). It’s essential to provide the right tools to match both the intended purpose and audience.

§ Achieve Critical Mass. Getting online discussions going requires active solicitation, facilitation and leadership.

§ Create an Economy. Successful online discussions can be seen as information economies in which reputation and attention are the forms of currency. People need to feel like they are rewarded/recognized for their contributions.

Some examples of sites with reputation management and reward systems include:

Slashdot http://slashdot.org/

Amazon http://www.amazon.com/exec/obidos/tg/cm/member-reviews/-/AFVQZQ8PW0L/102-8685923-0249765

Epinions http://www.epinions.com/

Database of Projects, Applications & Tools

This searchable database will provide descriptions of and pointers to projects, applications and tools related to end-to-end performance.

The modeling of objects and relationships will be coordinated with the directory of people and organizations and the digital library, enabling users to seamlessly search and browse across the following:

§ Projects

§ Applications

§ Tools

§ People (connection to directory of people and organizations)

§ Organizations (connection to directory of people and organizations)

§ Books, Articles, Web Sites (connection to digital library)

We will explore opportunities to partner with, leverage and learn from related directories such as the NLANR Advanced Applications Clearinghouse (http://dast.nlanr.net/Clearinghouse/).

Digital Library

The digital library provides a focal point for connecting people to content. Content will include: books, articles, web sites, seminars, conferences, archived netcasts, video clips, how to’s, reference documents, training materials, and more.

Success factors in developing the digital library include:

§ Collection Development Policy. Clearly defining what will and will not be included in terms of topic, ownership, format, quality, etc.

§ Local & Remote Resources. Defining policies and templates to handle local (Internet2) content and remote content. Catalog records with structured metadata and remote indexing will help to bridge this divide.

§ Organization & Labeling. A variety of organization schemes (e.g., geographic, network topography, hardware/software, subject) and a glossary of terms that maps technical language to that of the lay person will be needed.

Problems to Solutions

The following sections present 2 different ways to approach the same challenge: connecting end-to-end performance problems with real-world solutions. The best framework may be a hybrid of these approaches.

This area represents an important long-term goal of E2Epie. It’s also very complex and will require significant requirements definition and planning. We are particularly interested in feedback and suggestions regarding development of this component.

The KnowledgeBase

One model is that of a true KnowledgeBase (KB) as exemplified by Microsoft.

http://support.microsoft.com/directory/

This definition of KB centers around customer support, technical support, help desks and call centers. The notion is that a customer or user has a problem or question and the KB will provide a solution or answer. Natural language queries or step by step query builders are provided to assist the user in explaining or articulating their situation. So, there is typically a sophisticated front-end interface and a rich, structured back-end “knowledgebase” populated by a centralized, trained staff. Technologies for collaborative filtering can be part of this solution (e.g., top problems of the week/month, people who had that problem also had this problem).

This model will require centralized staffing (PERF) and/or strong coordination of distributed NOCs.

The ExpertConnection

This model represents a more distributed framework for addressing the problem resolution challenge.

The general idea is that each campus would have a designated Internet2 E2E Performance contact. Researchers and IT staff at that institution would be instructed to forward their E2E Performance problems to that person. This person would also be the designated contact for those outside that institution.

Upon receiving a problem, this E2E expert would perform some basic troubleshooting and analysis and then forward the problem onto the appropriate NOC. Incentives for people to become E2E experts might include:

§ Full access to the database of experts.

§ Access to performance measurement tools.

§ Access to a centralized Internet2 E2E Performance team (PERF).

These E2E experts could use tools and/or a process for trouble shooting, communication and problem resolution that results in the population over time of a rich online database or “knowledge base.” But this approach recognizes that the most valuable knowledge will continue to be that which resides in the heads of experts (and in their collective ability to work together).

Closing Statement

It is important to remember that our current definition of E2Epie is very much a work in progress. We are distributing this document early in the process as a catalyst to provoke feedback and discussion from the broader community of Internet2 members and partners. We very much want to hear from you. What do you expect from E2Epie? What advice do you have? What should we avoid? Who should we talk to?

Please direct your questions and suggestions to:

Cheryl Munn-Fremon

Director, Internet2 End-to-End Performance Initiative

cmfremon@internet2.edu, phone, address