David C. DiNucci
2004-07-17 11:12:08 UTC
I have intermittently posted an "Almost FAQ" here, adapted from the
original charter, to help guide the discussion on this group. It
apparently hasn't been very helpful, perhaps in part due to its
generality and/or political correctness. I'll be a bit more
focused/blunt here, so this should be taken as my personal, totally
unofficial statement.
Over the years, distributed computing, distributed systems, distributed
collectives, etc., have had many different flavors, have gone through
many hype cycles, and have encountered barriers on many fronts (e.g.
technical, political, economical, practical). At about the time that
this group was created, two "technologies" encountering these issues
were P2P and Grids. At least two groups had been formed to address
these issues: The Peer-to-peer working group (P2PWG), an industrial
group formed initially by Intel and other companies; and the Grid Forum,
formed initially by academic and government sources, notably sources
tied closely to the Globus project.
I attended meetings of both of these groups from time to time, including
some of the very first of each, as I had been working with "grid"
development at NASA since '96 or '97, and much longer in parallel and
distributed systems. Eventually, the Grid Forum swallowed up the P2PWG
as well as other distributed efforts (e.g. the New Productivity
Initiative), and the Grid Forum ultimately became the Global Grid Forum
(GGF), which continues to this day. Last I checked, hundreds of
representatives from government agencies and industry were flying all
over the world and paying sizeable registration fees every few months to
meet, and to talk about...well, largely Globus.
That, in itself, is not necessarily bad. What *is* bad is when people
are actively discouraged from discussing anything *but* Globus. I was,
at least twice, quite actively and rudely discouraged from using Grid
Forum mailing lists for un-Globus-ish discussions by the moderators of
those lists. These were not anti-Globus discussions, but just
suggestions by me that there might be more productive ways to handle
certain problems than the way Globus was handling them, or suggestions
that perhaps we should occasionally start with tabula rasa, rather than
with Globus, when determining solutions to certain issues. (I do believe
that the "power structure" of the GGF has changed markedly since that time.)
Feeling that the potential of these distributed systems was too great to
be left primarily in the hands of the GGF, my intent for this newsgroup
was to help foster a purer meritocracy when finding solutions to
distributed systems issues than I was (and am) seeing elsewhere. Raj
Buyya and I intentionally tried to use terms other than just P2P and
Grid to describe the scope of this newsgroup, in part because of the
ties these terms had to projects like Napster and Globus. The terms
Distributed (or Networked) Resource Collectives (or Cooperatives) were
some of our choices, but such terms can only enter the vernacular with
help from others.
So why is this group not thriving? Perhaps my experiences were very
different than others. Perhaps other forums (forii) and open source
projects more than adequately provide for everyone's needs in this
field. Maybe, even if they don't, the costs of appearing to publically
buck the system are too great. Maybe distributed technologies have lost
their luster, contrary to what the press is saying. And maybe, even if
they're still hot, people believe that their ideas are too valuable to
share here for free. Maybe the creation of this newsgroup was
wrong-headed from the beginning. (Based on the vote discussion from the
formation of this group, I'm guessing that some would find great
satisfaction if I was to concede such--and I'm not.)
Some of us have now had 8 years or so of exposure to "Grids", maybe 5 to
"P2P", and who knows how many to other forms of distributed resource
collectives. After that, aren't there some novel big picture ideas,
productive criticisms, etc., that can be shared on this group? As many
know, I've been working in one direction with Software Cabling, PICA
(People, Instruments, Computers, and Archives), and CDS (Cooperative
Data Sharing) for some time now, and I've shared many of my design
criteria, etc., on this group already. I'd be happy to continue, if
prompted, but even better would be to hear some comparisons, contrasts,
criticisms, etc., between different projects and approaches, including
new ideas.
-- Dave
-----------------------------------------------------------------
David C. DiNucci Elepar Tools for portable grid,
***@elepar.com http://www.elepar.com parallel, distributed, &
503-439-9431 Beaverton, OR 97006 peer-to-peer computing
original charter, to help guide the discussion on this group. It
apparently hasn't been very helpful, perhaps in part due to its
generality and/or political correctness. I'll be a bit more
focused/blunt here, so this should be taken as my personal, totally
unofficial statement.
Over the years, distributed computing, distributed systems, distributed
collectives, etc., have had many different flavors, have gone through
many hype cycles, and have encountered barriers on many fronts (e.g.
technical, political, economical, practical). At about the time that
this group was created, two "technologies" encountering these issues
were P2P and Grids. At least two groups had been formed to address
these issues: The Peer-to-peer working group (P2PWG), an industrial
group formed initially by Intel and other companies; and the Grid Forum,
formed initially by academic and government sources, notably sources
tied closely to the Globus project.
I attended meetings of both of these groups from time to time, including
some of the very first of each, as I had been working with "grid"
development at NASA since '96 or '97, and much longer in parallel and
distributed systems. Eventually, the Grid Forum swallowed up the P2PWG
as well as other distributed efforts (e.g. the New Productivity
Initiative), and the Grid Forum ultimately became the Global Grid Forum
(GGF), which continues to this day. Last I checked, hundreds of
representatives from government agencies and industry were flying all
over the world and paying sizeable registration fees every few months to
meet, and to talk about...well, largely Globus.
That, in itself, is not necessarily bad. What *is* bad is when people
are actively discouraged from discussing anything *but* Globus. I was,
at least twice, quite actively and rudely discouraged from using Grid
Forum mailing lists for un-Globus-ish discussions by the moderators of
those lists. These were not anti-Globus discussions, but just
suggestions by me that there might be more productive ways to handle
certain problems than the way Globus was handling them, or suggestions
that perhaps we should occasionally start with tabula rasa, rather than
with Globus, when determining solutions to certain issues. (I do believe
that the "power structure" of the GGF has changed markedly since that time.)
Feeling that the potential of these distributed systems was too great to
be left primarily in the hands of the GGF, my intent for this newsgroup
was to help foster a purer meritocracy when finding solutions to
distributed systems issues than I was (and am) seeing elsewhere. Raj
Buyya and I intentionally tried to use terms other than just P2P and
Grid to describe the scope of this newsgroup, in part because of the
ties these terms had to projects like Napster and Globus. The terms
Distributed (or Networked) Resource Collectives (or Cooperatives) were
some of our choices, but such terms can only enter the vernacular with
help from others.
So why is this group not thriving? Perhaps my experiences were very
different than others. Perhaps other forums (forii) and open source
projects more than adequately provide for everyone's needs in this
field. Maybe, even if they don't, the costs of appearing to publically
buck the system are too great. Maybe distributed technologies have lost
their luster, contrary to what the press is saying. And maybe, even if
they're still hot, people believe that their ideas are too valuable to
share here for free. Maybe the creation of this newsgroup was
wrong-headed from the beginning. (Based on the vote discussion from the
formation of this group, I'm guessing that some would find great
satisfaction if I was to concede such--and I'm not.)
Some of us have now had 8 years or so of exposure to "Grids", maybe 5 to
"P2P", and who knows how many to other forms of distributed resource
collectives. After that, aren't there some novel big picture ideas,
productive criticisms, etc., that can be shared on this group? As many
know, I've been working in one direction with Software Cabling, PICA
(People, Instruments, Computers, and Archives), and CDS (Cooperative
Data Sharing) for some time now, and I've shared many of my design
criteria, etc., on this group already. I'd be happy to continue, if
prompted, but even better would be to hear some comparisons, contrasts,
criticisms, etc., between different projects and approaches, including
new ideas.
-- Dave
-----------------------------------------------------------------
David C. DiNucci Elepar Tools for portable grid,
***@elepar.com http://www.elepar.com parallel, distributed, &
503-439-9431 Beaverton, OR 97006 peer-to-peer computing