[Oskar]

OSX Server has had out-of-the-box grid computing capabilities for a while, including the ability to connect to other computers over the internet for "@home" calculations ("cycle harvesting"). However, from my understanding, it only works with other macs.

SETI has a famous @home project that works with windows, mac, and linux. Is there any third-party grid computing software for OSX Server that works on them all, like SETI's project?

(I am aware of Oracle database 10g, but Apple seems to indicate that it's not for cycle harvesting).

Thanx!

[ghoti]

SETI@Home and a bunch of other projects now use the Berkeley Open Infrastructure for Network Computing (BOINC). There are a few others like Beowulf, though I don't know if any of those work on the Mac.

[Oskar]

Ah, I wasn't aware SETI used someone else's software. Thanks very much.

[Enki]

SETI is Berkeley. BOINC derived from code originally generated for the SETI project. See the paper that announced the start of BOINC: http://boinc.berkeley.edu/boinc2.pdf

[Oskar]

Does anyone know why apple doesn't port Xgrid to windows? I mean, who would want to make an @home project if they can't tap into over 95% of all computers?

[Batman]

Originally Posted by Oskar Does anyone know why apple doesn't port Xgrid to windows? I mean, who would want to make an @home project if they can't tap into over 95% of all computers?

Any computer can run "@home" projects through the app for that project, not just macs. Xgrid is for easily creating a grid accross a local network or the internet of macs, not PCs.

Personally, I have my family's windows box running Folding@home (a Stanford project about folding/misfolding proteins and finding cures to diseases) while idle. Personally, I find it much more usefull than trying to find aliens (even though that is cool too).

[Oskar]

Batman,

From the website:

Internet agents volunteer to donate cycles with large-scale “@Home” calculations.

It appears that Xgrid is capable of dealing with computers (err, macs) across the internet.

Edit: BTW, I am aware that any computer can run @home projects. But not if the grid software is Xgrid.

[Batman]

Originally Posted by Oskar ---snip--- Edit: BTW, I am aware that any computer can run @home projects. But not if the grid software is Xgrid.

That is why the large projects do not use Xgrid - they want to be able to tap into all the computers out there, not just Macs. Don't get me wrong, I'm a Mac user and love Xgrid, but I do't think it needs to be ported to Windows, nor do the iLife Apps (minus iTunes), or the Pro Tools.

[chucker]

I'm not sure where the conception that Xgrid is limited to Mac OS X comes from, because it's not. For once, the protocols used by Xgrid are, IIRC, fairly common amongst grid-like tools, though I could be wrong in that. Even so, however, there is, and has been for a long time (even before Apple's Xgrid reached 1.0) a portable Unix xgridagent. You still need to use OS X server to manage the grid, but that's hardly a show-stopper.

[Oskar]

To both of you, I need to make it clear that I don't want to be able to manage the grid on other OSes, I just want the client part to be available for everyone so I can tap into them.

Originally Posted by Batman Don't get me wrong, I'm a Mac user and love Xgrid, but I do't think it needs to be ported to Windows, nor do the iLife Apps (minus iTunes), or the Pro Tools.

That's a wrong comparison. Apple has no financial incentive to port iLife or the pro tools to Windows, because it would cut into its hardware sales. Porting the client side of Xgrid would help hardware sales, because it makes Xgrid more appealing, which makes OSX Server more appealing - and we all know that OSX Server needs Apple hardware to run it.

Originally Posted by chucker I'm not sure where the conception that Xgrid is limited to Mac OS X comes from, because it's not. For once, the protocols used by Xgrid are, IIRC, fairly common amongst grid-like tools, though I could be wrong in that. Even so, however, there is, and has been for a long time (even before Apple's Xgrid reached 1.0) a portable Unix xgridagent. You still need to use OS X server to manage the grid, but that's hardly a show-stopper.

Common protocols don't matter; the fact is, the client part of Xgrid is only available on the Mac. Indeed there are ports to allow Unix machines to play along, but I don't care about them. I care about the 95%+ out there running Windows.

[chucker]

Originally Posted by Oskar To both of you, I need to make it clear that I don't want to be able to manage the grid on other OSes, I just want the client part to be available for everyone so I can tap into them.

It is. I'm sure it's a non-issue to compile xgridagent on Windows through Cygwin or MinGW.

Porting the client side of Xgrid would help hardware sales, because it makes Xgrid more appealing, which makes OSX Server more appealing - and we all know that OSX Server needs Apple hardware to run it.

That would mainly boost sales of Xserves, which are already selling just fine. It wouldn't give them a killer app either, as other server platforms have grid computing systems of their own.

Common protocols don't matter; the fact is, the client part of Xgrid is only available on the Mac. Indeed there are ports to allow Unix machines to play along, but I don't care about them. I care about the 95%+ out there running Windows.

Windows for grid computing is like Mac OS X for gaming.

[Oskar]

Originally Posted by chucker It is. I'm sure it's a non-issue to compile xgridagent on Windows through Cygwin or MinGW.

Do you mean use an emulator? I'm not familiar with the things you mentioned.

Originally Posted by chucker That would mainly boost sales of Xserves, which are already selling just fine. It wouldn't give them a killer app either, as other server platforms have grid computing systems of their own.

The point of business is not to reach a point when you are selling just fine; the point is to maximize long-term revenue. I don't pretend that xserves will fly off the shelf as a result, but it could help.

Originally Posted by chucker Windows for grid computing is like Mac OS X for gaming.

SETI seems to be using Windows PCs for its project just fine.

[Enki]

SETI is not grid computing. SETI is distributed computing of completely non-interdependent packets of data. You might say it is plugged into a grid, but just using the pre-grid functionality of the internet. Still a very good and important capability, but fairly primitive all the same.

Yes, that description hinges on one particular definition of grid computing that sets the bar fairly high. But if the bar isn't set that high the term "grid computing" is meaningless because it would be just another catch-phrase that has no meat behind it.

[Oskar]

Originally Posted by Enki SETI is not grid computing. SETI is distributed computing of completely non-interdependent packets of data. You might say it is plugged into a grid, but just using the pre-grid functionality of the internet. Still a very good and important capability, but fairly primitive all the same. Yes, that description hinges on one particular definition of grid computing that sets the bar fairly high. But if the bar isn't set that high the term "grid computing" is meaningless because it would be just another catch-phrase that has no meat behind it.

I understand the distinction between local clusters and distributed computing. I figured Xgrid was the latter due to its across-the-internet @home capabilities. Please correct me if I'm wrong - I certainly am not an expert at this.

[chucker]

Originally Posted by Enki SETI is not grid computing.

Wikipedia would disagree.

SETI@home ("SETI at home") is a grid computing (distributed computing in the project's own terminology) project using Internet-connected computers [..]

SETI is distributed computing of completely non-interdependent packets of data.

Which is grid computing. If the packages were interdependent, you'd have a fairly inefficient grid.

Yes, that description hinges on one particular definition of grid computing that sets the bar fairly high. But if the bar isn't set that high the term "grid computing" is meaningless because it would be just another catch-phrase that has no meat behind it.

Grid computing is much like cluster computing, except it adds more flexibility. Grid nodes don't need to be similar in specs, don't need to be always available, don't need to frequently sync with each other, don't need to use the same client and aren't usually stored at the same facility.

[Oskar]

I've found a pretty interesting Xgrid agent that may alleviate the problem. It's a java app, so I guess it would work on Windows. I don't know if it's any good, but at least I know it's possible.

[Enki]

Chucker,

The problem with accepting one of Wikipedia's statements that SETI@home is grid computing is that it is contradicted by the first sentence of the actual grid computing entry. Does that make it authoritative that SETI isn't grid computing? No, probably just a coincidence that happens to agree with me - read on.

"The Origins" part of grid computing gushes about SETI! Maybe I'm all wet??? Maybe SETI@home is a grid computing project??? Wait a second, how come all the papers mentioned in the "Definitions of Grid Computing" section have definitions that don't include how SETI@home is deployed? How contradictory can we get? Wikipedia can't seem to make up it's collective mind! The last line here is perfect though "Grid Computing is a super set of distributed computing."

The "Historical Sweep" in the "Conceptual Framework" subsection of grid computing has it exactly right and actually fits with the last line in the definitions section. SETI is mentioned as an important distributed computing project that used idle CPU cycles. SETI is an excellent example of distributed computing that served as an example that a grid may have a real future, even though it is not truly a grid computing project itself.

I have found Wikipedia to be close to 50% technically correct on in-depth CS technical topics that I feel qualified enough to judge. A perfectly crappy record and one to be expected given anyone can get permission to wiki. It's a great place to start, but too uneven to be taken seriously as a last word. Especially given the all too prevalent internal contradictions in many of the entries that have more than one contributor.

[my gawd that read like horsecrap the first time around]

[BarracksSi]

I do not consider SETI@Home to be grid computing.

If I were to network my two iBooks together and run XGrid on them, then that's grid computing. That would be my own little supercomputer cluster.

SETI@Home, at its most basic form, is like me downloading an image, resizing it, and sending back the new version. That's all it is.

If there are more technical explanations out there that blur the distinction between grid computing and distributed computing, I really don't care.