The Apache Group
A Case Study of Internet Collaboration
and Virtual Communities

Roy T. Fielding
Dept. of Information & Computer Science
University of California, Irvine

• Virtual Communities

• What is Apache?
  • Founders
  • Current Apache Group
  • Development Context
  • Secrets of Our Success

• Collaborative Development
  • Impact of the Internet on Collaboration
  • Collaboration within the Apache Group

• The Problems with Being Virtual

Virtual Communities

• "Virtual" == "Not"?
  • "Virtual" == "Cyberspace" (another loathsome term)
  • A community that depends upon cyberspace for interaction

• What, then, is a Community?
  • Webster has 12 definitions, all poor
  • My definition:
    A group of organisms bound together by enlightened self-interest and characterized by a common product

• Virtual Communities
  • A newsgroup is not, in and of itself, a virtual community
  • The Internet Community (R.I.P.)
  • The Apache Group

What is Apache?

• Formed in February `95
• 8 Founding Members (all end-users)

• To provide a public, secure, efficient and extensible server which provides HTTP services in sync with the standards.

• #1 Server (46% market share / 517,000 sites)
• ~12 active developers (5 of the founders)
• Hundreds of contributors

Apache Group Founders, Feb 1995

	Brian Behlendorf	HotWired, Bay Area, California
	Roy Fielding	UC Irvine Ph.D. Student, California
	Rob Hartill	LANL, New Mexico
	David Robinson	Cambridge Ph.D. Student, UK
	Cliff Skolnick	Sun Microsystems, California
	Randy Terbush	Zyzzyva ISP, Nebraska
	Robert Thau	MIT Ph.D. Student, Massachusetts
	Andrew Wilson	Elsevier, Oxford, UK

indicates people Roy has met in person

Initial contributors included Eric Hagberg (Cornell Medical College), Frank Peters (Mississippi State University), and Nicolas Pioch (The Unofficial Louvre, Paris, France).

Current Apache Group, May 1997

	Brian Behlendorf	Organic Online, California
	Ken Coar	Process Software Corp., New England, USA
	Mark Cox	UKWeb Ltd., UK
	Roy Fielding	UC Irvine Ph.D. Student, California
	Dean Gaudet	Steam Tunnel Operations, California
	Rob Hartill	Internet Movie DB, UK
	Jim Jagielski	jaguNET ISP, Maryland
	Alexei Kosut	Senior in High School, California
	Ben Laurie	Freelance Consultant, UK
	Chuck Murcko	The Topsail Group, Pennsylvania
	Aram Mirzadeh	Qosina Corporation, New York
	Sameer Parekh	C2Net, California
	Paul Sutton	UKWeb Ltd., UK
	Marc Slemko	Edmonton, Canada
	Randy Terbush	Zyzzyva ISP, Nebraska
	Dirk-Willem van Gulik	Freelance Consultant, Italy
	Andrew Wilson	Freelance Consultant, UK

indicates people Roy has met in person

There are many more Regular Contributors to the Apache Project.

Development Context

• USA, UK, Canada, Italy, The Netherlands, ...

• Everybody has (at least) one other real job

• Enlightened Self-interest

• Coordination tasks are voluntary
• Volunteers get tired/busy, rotate tasks
• Decisions made by consensus, voting
• 36 hours to make a serious decision

Secrets of Our Success

• The End-Users are the Developers
• Core developers act as Quality Control
• System designed for Extensibility (API, source code)

• Public support line (newsgroup, bugdb)
• Problem -> Patch, typically less than a week
• Web-based Distributions

• Unlimited supply of ideas (good and bad)

Collaborative Development

Requires:

• Warm Bodies

• At least one common goal
  • but not all goals are held in common

• Communication
  • Public: one to all, recorded for later review
  • Private: one to few

• Shared Information Space
  • Access to past communication
  • Access to past and current development products

• Coordination
  • to make all of the above possible

Impact of the Internet on Collaboration

• Millions of Warm Bodies to choose from!

• Many goals held in common
  • Easy to find people when working on a global scale

• E-mail for Communication
  • Public: mailing list, archived in Hypermail
  • Private: off-list mail

• The Web is a Shared Information Space
  • Access to past communication
  • Access to past and current development products

• Coordination
  • ... oops, still a work in progress

Collaboration within the Apache Group

• E-mail Communication
  • Developer mailing list (avg. 50 msgs/day)
  • Private mail among core developers

• Shared Information Space
  • Machine in Bay Area
    • FTP, HTTP, and remote CVS servers
    • Logins for core developers
  • Public area for distributions,
    Developer area for work-in-progress

• Coordination
  • Voting (on patches, releases, strategy, etc.)
  • Agenda

Process Evolution

1995 Feb

Patches exchanged by mail, given a name,

vote tabulated (+1 yes, -1 veto, +0 abstain)

patches applied,

generate new release

1995 Mar

Formal patch database/forms tried, later replaced by simple directory

1995 Sep

Voting guidelines codified,

separate roles of voter, vote coordinator, release builder

1996 Jan

Remote CVS enabled

1996 Nov

GNATS problem tracking system

1997 Jan

Voting restored

Agenda created

The Problems with Being Virtual

• Limited Support for Coordination
  • Keeping track of progress and potential conflict has required a deluge of communication (e-mail)
  • Maintaining the Agenda is a manual process

• Entry Barriers
  • Difficult to browse the current software state
  • Difficult to separate areas of development

• Apache succeeds in large part due to the unusual expertise of its user/developers
  • all Internet experts
  • capable of compiling, installing software

• E-mail is not sufficient for resolving interpersonal conflicts that are more than just "differences in opinion"