Path: gmdzi!unido!mcsun!uunet!ogicse!uwm.edu!bionet!agate!darkstar!osf.org!
steiner
From: stei...@osf.org
Newsgroups: comp.os.research
Subject: OSF description
Message-ID: <1523@darkstar.ucsc.edu>
Date: 28 Feb 90 15:29:35 GMT
Sender: use...@darkstar.ucsc.edu
Organization: Open Software Foundation, Cambridge MA
Lines: 256
Approved: comp-os-resea...@jupiter.ucsc.edu
Posted: Wed Feb 28 16:29:35 1990
Hi Darrell,
In response to your invitation, below is a description of OSF and some
of its activities. There are three main sections, one on OSF in general,
the second on the Distributed Computing Environment RFT project, and the
third is on the OSF Research Institute.
There are many other activities going on at OSF; this article just
describes the work I happen to be involved with here, and aspects
of it that I think might be interesting to comp.os.research readers.
Jennifer Steiner
---------------------------------------------------------------------------
OSF
===
The Open Software Foundation is a non-profit international organization
that acquires and offers systems software (source code, specifications,
verifications, and reference implementations) according to the needs of
its members, and through them the industry in general. Members include
OSF's sponsors (Apollo, Bull, DEC, HP, Hitachi, IBM, Nixdorf, Philips,
and Siemens), and members representing independent software vendors,
end-users, government agencies, research labs, and academic institutions.
There are currently around 180 members. They give input to OSF through
Member Meetings (2 or 3 are held per year), participation in Special
Interest Groups, responding to OSF surveys, etc. OSF, in turn, is a
member of standards organizations such as X/Open.
To date, OSF has produced one offering, the OSF/Motif User Environment
Component, and is working on a second, the OSF/1 operating system, based
on CMU's Mach 2.5 operating system. OSF/1 snapshots are currently
available to OSF's membership, and the system will be generally available
in November of 1990.
OSF has ~230 employees, and offices in Cambridge, San Francisco,
Washington DC, Munich, Tokyo, and Grenoble. The two main groups of
interest to this readership are Development and the Research Institute,
and some of their activities will be described below. Other departments
are human resources, operations, and finance.
One of the purposes of OSF is to provide a common software base for
open systems. One of the ways it does this is by examining different
technologies available in a given area (such as User Interfaces) and
choosing what it believes is the "best" and offering that to its membership
and the industry. The membership is large, so adoption of OSF's offerings
by the membership helps promote standardization. In addition, in evaluating
technologies, OSF factors in the work of standards bodies (e.g., ISO,
POSIX) and the existence of de facto industry standards (e.g., X and NFS).
Two important priorities for OSF are providing offerings that 1) support
interoperability and 2) are portable to different hardware, OS, and
networking environments.
RFT
===
The role of the Development organization at OSF is somewhat different
from that in a typical software company, because much of OSF's software
is acquired from technology suppliers rather than developed from scratch.
An example of the way OSF Development is unusual is the process it has
developed for acquiring software: the "Request For Technology", or RFT.
The RFT process was used to select the technology making up the current
Motif offering, and two RFTs are currently in progress: the Distributed
Computing Environment (DCE) RFT, and the Architecture Neutral Definition
Format (ANDF) RFT. I'll describe the DCE RFT in some detail below, since
that's the one I'm currently participating in. ANDF is exploring means
of distributing software securely in an architecturally-independent
manner (something like "shrink-wrap" software for UNIX).
DCE
===
The Distributed Computing Environment RFT was initiated because
of two indications from the OSF membership: The first was the result
of a member survey, which indicated that lack of interoperability
was the most critical problem facing the computer industry today.
The second was a recommendation by the Distributed Application
Environment SIG, a special interest group made up of interested OSF
members.
The RFT was issued in June of 1989. It called for technology making
up a "vendor neutral networking and distributed computing environment to
support distributed applications", soliciting "core services" such as RPC
and Presentation Services, Name/Directory Service, Authentication, and
Distributed File Systems. It also outlined some key evaluation criteria.
We received 49 Letters of Intent (preliminary descriptions of technology
to be offered). The technologies described in these letters fell into
17 different categories, 5 of which were considered beyond the scope of
the RFT (the scope was eventually narrowed even further). The remaining
responders were encouraged to send full submissions, and by the deadline
in October, we had received 29 of them.
Since then we have been reviewing the full submissions, gathering
feedback on them from our membership and consultants (see below),
determining a framework for distributed computing at OSF, and deciding
which areas belong within the scope of the current RFT. Right now
we are in a "laboratory evaluation" phase, in which we are
inviting submitters whose technology is in scope to come to
OSF and demonstrate and describe in detail their technology.
We plan to announce a decision in the second quarter of this
year, and to have an offering by the end of 1990.
We expect the Distributed Computing Environment offering to have the
following components: RPC and Presentation Service, Name Service,
Authentication Service, Time Service, Distributed File Service, and
Threads; and possibly PC Integration and Authorization.
A note of interest to the present readers, and for me one of the most
rewarding aspects of working on this project, is the list of OSF staff
and outside consultants who make up the DCE evaluation team.
The OSF staff are:
Dietmar Fauth (on sabbatical from Siemens)
Jon Gossels (Business Area Manager for Interoperability)
Doug Hartman (Director of RFTs)
Brad Johnson (on sabbatical from DEC)
Ram Kumar (DCE Technology Manager)
Norbert Leser
Dick Mackey
Chi Shue
Todd Smith (of Tivoli Systems)
Jennifer Steiner (OSF RI)
Walt Tuvell
The outside consultants are:
Andrew Birrell, DEC SRC
Heinz-Juergen Burkhardt, GMD
David Cheriton, Stanford University
Paul Mockapetris, USC ISI
Sape Mullender, CWI, the Netherlands
Roger Needham, U. of Cambridge Computer Laboratory
William Pigott, DHL Worldwide Express
Rusty Sandberg, Legato Systems, Inc.
Peter Schay, Gartner Group, Inc.
Walter Ulrich, Arthur D. Little, Inc.
Peter J. Weinberger, AT&T Bell Labs
Research Institute
==================
The OSF Research Institute has two offices - one in the Cambridge
headquarters, and another in Grenoble, France. The RI currently
has around 25 employees, and will grow to around 35 by the end of 1990.
The Research Institute has several areas of activities. These can be
described first by the RI's Mission statement, and secondly by the
concrete programs that are currently in place.
RI Mission
1) Involve the research community worldwide in the evolution of open
systems
- Recruit members from the research community - academic,
industrial & government
- Communicate their needs to OSF and communicate the needs of
Open Systems to them
- Conduct workshops and publish proceedings on areas of special
interest
- Encourage acceptance and early adoption of OSF technologies
- Facilitate research community involvement in the Open Process
2) Analyze emerging technologies to identify opportunities for future
RFTs or OSF development
- Fund key areas of research critical to the progress of Open
Systems
- Collaborate in critical areas so as to have a deep
understanding of the technology
- Test, measure, integrate, extend and disseminate emerging
technologies
- Plan for backward compatibility
- Communicate technical opportunities to OSF membership
3) Build a long-term shared vision and architecture for the future of
Open Systems
- Identify major architecture and technology trends
- Publish architectural specifications and vision papers for
review
- Manage the process of building a roadmap for OSF's technology
directions
The RI Today
The above mission statements have been materialized in the following
ways.
1) Research relations: interfacing with academic, industry research,
and government OSF members (and potential members), holding workshops,
funding research projects outside of OSF, prototype distribution,
and technical support. Approximately 1/3 of OSF's members are
research or educational institutions.
- Workshops held to date:
User Interface workshop (April 89)
Mach workshop (Nov 89)
- Informational services:
Videotape series on Mach tutorials and workshop
Bibliographies on Mach
- Distribution of research results
MIT Project Athena's Motif project (available now)
Univ. Guelph Motif Graphical shell (available March)
Univ. Lowell Motif language bindings (available March)
- Grants were awarded in 1989 to 14 institutions, mostly
universities, for work such as Motif, Unix, and
micro-kernel development.
- The Research Institute plans to publish a quarterly newsletter,
with the first issue appearing in March of this year.
2) Advanced Development
RI Advanced Development is currently focusing on development of a prototype
micro-kernel operating system, using Mach 3.0 from CMU and other distributed
operating system technologies as bases. We expect that the results of our
efforts will be taken by OSF Development as input for their continuing
development of the OSF operating system offering, OSF/1.
We will be gathering input for this project from the research community in
several forms, such as sabbaticals at OSF, advisory consultation, work done
under grants and subcontracts, direct acquisition of source technology, and
research results published in the literature.
Some known goals of the program are to provide a base
- for a B3 secure system
- that can offer multiple environments (flavors of
Unix, DOS, etc.)
- that is extensible (able to support new types of
kernel-level services, e.g. database servers)
- useful for innovation by CS researchers in a variety of
endeavors, such as imaging and object technology
while keeping in mind that we hope OSF Development will profit from
our work, which means it would ideally be binary backward compatible
with OSF/1 with a minimum performance penalty.
3) Working with OSF Development
Helping identify areas appropriate for RFTs
Contributing RI personnel to RFT evaluations
(1 person is currently on loan to Development for
the DCE RFT - me)
Helping identify potential outside consultants
(e.g., for RFT evaluations)
4) Long-term direction for OSF
Various projects are being considered, including:
White Paper on Distributed Operating Systems (in progress)
Book on Micro-kernel Technology (under consideration)