Red Hat has begun selling new software for creating a communal storage system that spans many devices in a network–part of its effort to expand beyond its core operating system business.
Red Hat unveils storage software
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Kudos to RedHat for choosing to release this new filesystem under the GPL continuing their non-proprietary traditions. Sweet.
The GFS filesystem is now available for download at http://sources.redhat.com/cluster/
According to a post made to the Linux Kernel Mailing List, Red Hat hopes to get the filesystem included into the standard kernel in the future. It would be the first clustered filesystem in the kernel and a big boost for Linux in general IMO.
” It would be the first clustered filesystem in the kernel and a big boost for Linux in general IMO.”
lustre is already schedules for inclusion and some specific patches readying it has already been included in the last kernel release. lustre is already sponspored by HP and cluster filesystems and is probably used as much as gfs. so it wont be the only cluster filesystem in linux
incidentally the intermezzo filesystem was already in the kernel but was removed in the last release due to non maintainance (it was designed by the same developer as lustre)
I was under the impression only the VFS changes from Lustre would be merged so that vendors or end-users could use out-of-kernel compiled Lustre modules without needing to recompile or otherwise modify their kernels.
But I’m all for including Lustre as well.
“I was under the impression only the VFS changes from Lustre would be merged so that vendors or end-users could use out-of-kernel compiled Lustre modules without needing to recompile or otherwise modify their kernels.
”
the current kernel release already makes this possible but i believe that since intermezzo was in the kernel before and this is the same guy he would want to include lustre within the vanilla sources. if people can just install another package and get lustre running without modifying/patching the kernel sources i dont think any user would bother to notice the difference depending on how well the distro integrates the stuff.
if you have anything to suggest that lustre wouldnt be included i would like to know more abt it. any links?
This is how Redhat got into the position it is in and why it will STAY in it’s current position. Nothing at ALL beats having the goodwill of your current and possible customers.
This is why I stick with Fedora – because Redhat has my goodwill and loyality by their business model.
I’m almost certain the current Lustre release does not work with any official unpatched 2.6 kernel at the moment. The proposed Lustre VFS changes have not yet been merged and it’s been pretty quiet on that front lately.
Here’s the recent LKML thread about merging the VFS changes that would allow Lustre to work without patches: http://marc.theaimsgroup.com/?l=linux-kernel&m=108539900503192&w=2
Certainly this patch, if accepted, also makes a possible inclusion of Lustre easier in the future, but as I said I haven’t seen that proposed.
“http://lwn.net/Articles/87844/“
yes. i read about this before and somehow got the impression that it was already in because the editor concludes that its likely to go in
Go Redhat!!! Gooooooooo!!!!!!!!!!!!!!!!
whenever i am in the position to get redhat business (and i mean paid as well as just end users using it supporting it) i always do.
except for their seeming hatred of kde (and i can understand why after pumping lots of resources into gnome) they do things the right way most of the time IMO
The rise in importance of proprietary identity management tools has come about because of a shortcoming in the Unix architecture: although it is a multiuser system, it has historically been single-machine. Since nearly all computing environments of any significant size are multimachine, this has led to a nontrivial amount of grief, and spit-and-glue hack solutions to avoid it.
The goal, therefore, has to be making entire networks as easy to navigate as one machine – kind of like Plan 9 did years ago, but with backwards compatibility. These open-source clustering technologies from HP and Red Hat are an important step toward that goal. Unlike their expensive proprietary cousins, they are likely to catch on, much as the de-facto open-source Unix architecture stuck around while its technically more advanced but proprietary cousin VMS has passed away.
I agree. Besides the consistent quality and testedness of the package they put out, the reason our shop still buys Red Hat for our servers is that since they are a pure-play OSS shop, we know they aren’t trying to bait-and-switch us with proprietary management apps layered over the free base. (Unlike certain other big Linux vendors).
We’re willing to swallow the price tag because of that trust. In fact, this might be the most important thing OSS brings to the software market: trust. I know it’s much more important in my father’s auto parts business than it is in most of the computing concerns I’ve worked at. Perhaps IT is becoming a normal field.
>>technically more advanced but proprietary cousin VMS has passed away.
Last time i looked OpenVMS was alive and kicking and was booting on Itanium