HP, IBM, Intel and Sun Microsystems Launch Globus Consortium

Posted Jan 25, 2005 6:52 UTC (Tue) by jd (guest, #26381) [Link]

There is little doubt in my mind that Globus is a powerful toolkit. Powerful, but limited. It supports the original MPICH toolkit, but as far as I can tell, that is the only MPI support it has. Oh, MPICH is good, don't get me wrong, it just isn't good for everything. It doesn't do robust connections, MPI-2 or SCI networks, for example. I'm pretty certain it does not support OpenMP, either, which means you don't get optimized parallelization on multi-processor boards. Also, the bindings seem limited. There's Java, C (or perhaps C++) and I think Fortran. None of these are exactly designed for grid computing, which makes it much harder to design anything to put on a grid.

Those are the limitations I've found, and they're significant. They need to be fixed, if the scientific community is to make the best use of what Globus offers. Or, indeed, any use at all - scientists will make use of technology, but are notoriously resistant to new concepts. That is one reason Fortran is still used. If Globus - or grid computing in general - is to be attractive, it has to offer benefits that are glaringly obvious, even when it is outside the specialist field of those concerned. Physicists in Australia were running particle accelerators off PDP-11s in the mid 1990s - I know, I helped program them. These aren't people who embrace change for the sake of it.

As for the consortium itself - the four parties involved are known to be fiercely antagonistic towards each other. I hope this is a sign that there are people in each company with some common sense, but it would not be wise to rely on that too much. Of the four, IBM is the most open to the Free/Open Source community, with HP somewhere in the running. (Although HP has pulled many of its Linux projects, over time.) Intel grudgingly gave help to developers porting the Linux kernel to the Itanium, but I have heard nothing that suggests their support was either as early or as extensive as that to proprietary developers. Sun withdrew the Java license from FreeBSD developers, have spent a lot of time trying to bad-mouth Linux and have been pushing their semi-open licenses as being totally open and unencumbering. The four parties are worlds apart from each other, on the political and philosophical fronts. I give it a year, tops, before fist-fights break out.

Globus is a good product and it'll undoubtedly take grid computing from a mere abstract concept into something that is used in a day-to-day environment. It probably already has. I tend to look at Globus in the same way I look at PVM - an excellent solution for specific problems, one of the earliest prototypes and one that isn't likely to be totally replaced any time soon. It needs developers who can remain focused on their work, rather than avoiding knives in the dark.

Posted Jan 25, 2005 10:30 UTC (Tue) by danielos (subscriber, #6053) [Link]

Interesting:

http://www-unix.globus.org/toolkit/license.html

This is not listed in opensource.org, but I think this project will survive as is (free), and (giving Globus Consortium) it'll not fork.

Posted Jan 25, 2005 13:59 UTC (Tue) by NZheretic (guest, #409) [Link]

Take a pinch of Standard Linux
Wrap it up in Xen
Add a touch of SELinux
And a little bitty bit of Globus
Oh like a Sandboxed Platform
Oh Lordy, Lordy, mixed with Free and Open Source Code
You know you lump it all together
And you got a recipe for a Multi Vendor Development scene
It is coming though, you know, you know.

What we have is a great big melting pot
Big enough enough enough to take every vendor and all IT's got
And keep it stirring for a hundred years or more
And turn out Application Service and Content Providers by the score.

With apologies to Blue Mink.